Abstract
An adaptive technique for detection of ground stationary targets in a variable clutter environment by an airborne millimeter wave radar is presented. The scheme consists of three basic units: an adaptive decorrelator, a clutter classifier, and a constant false alarm rate (CFAR) processor. Due to the inherent nature of radar returns (echoes), the signal consisting of target(S), clutter(C), and noise(N) are highly correlated. These signals are passed through a whitening matched filters combination. The selection of the parameters of whitening and matched filters is performed through an adaptive identification scheme (Kalman algorithm) using autoregressive(AR) and autoregressive-moving average(ARMA) models. The output of the matched filter is then passed through the pattern classifier unit, which determines whether the given clutter sample belongs to the Log-normal type or Weibull type statistical distribution families, and determines the appropriate detection threshold for the CFAR processor unit.